dinov2-small-imagenet1k-1-layer-finetuned-face-match-4360

This model is a fine-tuned version of facebook/dinov2-small-imagenet1k-1-layer on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.3400
• Accuracy: 0.8463

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5925	1.0	109	0.7429	0.7523
0.4436	2.0	218	0.4149	0.8200
0.4574	3.0	327	0.4653	0.7787
0.4004	4.0	436	0.3668	0.8291
0.4814	5.0	545	0.4562	0.7626
0.437	6.0	654	0.3939	0.8257
0.3617	7.0	763	0.3683	0.8337
0.415	8.0	872	0.3529	0.8406
0.3611	9.0	981	0.3575	0.8337
0.4122	10.0	1090	0.3706	0.8268
0.3767	11.0	1199	0.3406	0.8291
0.3295	12.0	1308	0.3456	0.8360
0.3467	13.0	1417	0.3310	0.8521
0.2992	14.0	1526	0.3541	0.8291
0.2879	15.0	1635	0.3253	0.8509

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2